Landing AI

What it is for

Landing AI ADE is a document-to-structure API: you send a PDF or image, Parse returns grounded Markdown/JSON, then Extract / Classify / Split / Section do one job each. Use it when the job is credits-metered parsing you can cite on the page, not when you need a $20 chat seat.

What makes it unique: the cover is the official landing-ai/ade-python GitHub file list — Apache-2.0 ADE client, About “Agentic Document Extraction” — not a pricing card.

Key features

• Parse — convert a document into structured Markdown + hierarchical JSON with element chunks (text, tables, forms, barcodes) (ADE overview). The required first step. DPT-3 Pro vs Verity, and Standard vs Priority, are SKU of this call.
• Extract — pull schema fields from an already-parsed document (overview Extract). A key-value job, not a parse. Schema length changes the credit burn.
• Classify — label each page with classes you define, without Parse (overview). A router, not an extractor.
• Split — explode one file that contains many documents into sub-documents (overview). A separator, not a page label.
• Section — build a hierarchical table of contents with chunk references after Parse (overview). Navigation, not field extraction.

Visual grounding (page + coordinates) is output of Parse, not a sixth product. The Playground is ops UI for the same APIs. ade-python is install. Credits are the meter. Zero-data-retention and HIPAA BAA are Team/Enterprise SKUs.

Who it's for

Best fit: doc-pipeline teams who will burn the 1,000 free credits, then buy $1 = 100 credits (Team from $250/month when they need shared usage and ZDR).

Pricing

Verified 2026-10-22 (Asia/Shanghai) from landing.ai/pricing (Agentic APIs). Billing is credits. $1 buys 100 credits.

• Explore (pay-as-you-go) — start with 1,000 free credits. Then $1 / 100 credits. Parse, Extract, split/classify/section. No Team management.
• Team — from $250/month. Plans from 25,000 to 200,000 credits/month. Shared usage, email support, ZDR available, HIPAA with BAA.
• Enterprise — custom. SaaS / VPC / on-prem, SLAs, priority rate limits. For more than 200,000 credits/month.
• Credit burn (public FAQ) — typical business page (~4,000 characters) on Standard: DPT-3 Pro Parse ≈ 1.5 credits; DPT-3 Verity ≈ 0.55 credits. Extract v2 ≈ 1.5 credits per typical page. Priority is 2× Standard. Official formulas: Credit consumption.
• Marketplace: AWS and Snowflake are listed as purchase doors — confirm those SKUs there, not here.

Confirm checkout. Taxes follow your location.

Limitations / costs

• Free is 1,000 credits, then $1 / 100. Priority + Pro can cost 75–90% more than Verity + Standard on the same page.
• Playground is not a fifth API. It runs Parse/Extract you already pay for.
• A document API, not a warehouse GenBI desk. For text-to-SQL see Wren AI. For another parse vendor see Extend.

Alternatives

• Extend — another document-extraction API when you want a different parser.
• Unstructured — OSS + cloud partitioning when you will run the ETL yourself.
• Firecrawl — web-to-markdown when the source is a URL, not a PDF pack.

Official tutorial

Official LandingAI channel video: Playground vs. API: Run the Same Parse Both Ways (oEmbed author: LandingAI). First-party Parse in Playground and via API.

Written path: ADE overview plus Pricing and Credit consumption.

Website

https://landing.ai

Repository

https://github.com/landing-ai/ade-python

Source

Catalog row curated for AI Tools Nav. Cover is the official landing-ai/ade-python GitHub file list (ADE Python client). Dollars checked on landing.ai/pricing ($1 = 100 credits; 1,000 free; Team from $250/mo). Tutorial oEmbed author: LandingAI.

Last verified

2026-10-22 (Asia/Shanghai) editorial check.